Radiology · Pulmonary nodule malignancy risk stratification (CADx)
Optellum Virtual Nodule Clinic
Optellum Ltd.
The reference case for putting a number on a lung nodule. Optellum's Lung Cancer Prediction score out-discriminated the established Brock model in a multi-centre external validation and improved reader agreement in a Radiology reader study — and it was the first nodule-CADx cleared across the US, EU and UK at once. What the record does not yet hold is a prospective outcome trial, and much of the founding evidence carries manufacturer authorship.
Performance Metrics
Clinical Evidence
The evidence for the Virtual Nodule Clinic is about discrimination — how well the Lung Cancer Prediction score separates benign from malignant nodules — and it should be read in that frame, because the score's accuracy is well-characterised while its effect on patient outcomes is not yet tested. The founding study (Massion et al., American Journal of Respiratory and Critical Care Medicine 2020) trained the Lung Cancer Prediction CNN on National Lung Screening Trial images and tested it on incidental-nodule cohorts from Vanderbilt and Oxford, reaching areas under the curve of 0.92, 0.84 and 0.92 across the training and two validation sets and out-performing the clinical Mayo malignancy model on both external cohorts. The IDEAL-study external validation (Baldwin et al., Thorax 2020), run across Oxford, Nottingham and Leeds teaching hospitals, put the score's area under the curve at 0.896 against 0.868 for the Brock University model recommended in UK guidelines — a modest but statistically significant advantage (p ≤ 0.005), with the score identifying a larger fraction of benign nodules without missing cancers. Two things temper the picture. The strongest evidence for clinical effect is a multi-reader study (Kim et al., Radiology 2022), in which access to the score improved radiologists' specificity and their agreement on both risk category and management recommendation — a reader-behaviour result on a retrospective case set, not a prospective trial of downstream outcomes. And the founding literature is substantially manufacturer-affiliated: several authors on the Thorax and Radiology papers are Optellum employees, and the work grew out of the company's own development programme. A longitudinal extension of the model incorporating change across serial scans (Scientific Reports 2023) reported further gains but remains developmental. The honest summary: good, partly independent evidence that the score discriminates malignancy and shifts reader decisions, and no published randomised or prospective outcome data.
| Study | Design | n | Sensitivity | Specificity | AUC | Published |
|---|---|---|---|---|---|---|
| Massion PP, Antic S, Ather S, et al. (Optellum-affiliated development study) | RetrospectiveRetrospective | 0 | — | — | 0.92 / 0.84 / 0.92 (NLST / Vanderbilt / Oxford) | Am J Respir Crit Care Med 2020;202(2):241-249 (PMID 32326730); LCP-CNN trained on NLST, externally tested on incidental-nodule cohorts, out-performed the clinical Mayo model on both validation sets |
| Baldwin DR, Gleeson FV, et al. (IDEAL study; several authors Optellum-employed) | RetrospectiveRetrospective | 1,397 | — | — | 0.896 (95% CI 0.876–0.915) vs Brock 0.868 (0.843–0.891) | Thorax 2020;75(4):306-312 (PMID 32139611); multi-centre external validation across Oxford, Nottingham and Leeds; LCP-CNN discrimination significantly above the Brock model (p≤0.005) |
| Kim RY, Oke JL, Pickup LC, ... Vachani A (multi-reader study; Optellum-affiliated co-authors) | RetrospectiveRetrospective | 300 | — | improved with CAD | — | Radiology 2022;304(3):683-691 (PMID 35608448); computer-assisted use of the score improved reader specificity and interobserver agreement on risk category and management, on a retrospective case set |
| Longitudinal LCP-CNN model (Optellum-affiliated; developmental) | RetrospectiveRetrospective | 0 | — | — | — | Sci Rep 2023;13:6157 (PMID 37061539); a longitudinal model incorporating change across serial scans improved classification of indeterminate nodules over the single-timepoint score — developmental, not a clinical outcome study |
Clinical Pulse
Verified clinician sentiment
Verified clinician reviews launching soon · Apply to contribute →
Reviews are signed by clinicians verified against their regulator's public register.
Inside the algorithm
Editorial featureHow the Virtual Nodule Clinic scores a nodule.
Five stages — from raw input to verdict — drawn from manufacturer documentation and the public regulatory record.
- INGEST
- NORMALISE
- DETECT
- LOCALISE
- VERDICT
Stage 01 · INGEST
A chest CT study arrives, with a nodule already identified.
The software receives the chest CT study as objects from PACS or the imaging gateway and works on nodules that have already been found — by a radiologist, a screening read or a detection tool. It characterises an identified lesion rather than searching the whole scan for missed ones; detection and characterisation are separate jobs, and this device does the second.
Input
Chest CT (DICOM)
Inference
Per-nodule
Inside the Auris+ Listing
Five more sections complete this device’s Auris+ Listing.
Decision Ledger
ProPro unlocks a private, cross-vendor log of every case you read on this device — what the AI called, what you concluded, and a one-line reason if you overrode it. Ready for the EU AI Act's deployer logging obligations when they land in 2028.
Clinical Evidence Deep Dive
ProPro unlocks the structured clinical-evidence summary — study count, target patient population, and a tabular accuracy-metrics view drawn from peer-reviewed sources.
Peer-Reviewed Publications
ProPro unlocks the curated peer-reviewed publication list with PubMed cross-links — the citation backbone of every editorial verdict.
Post-Market & Regulatory Conditions
ProPro unlocks the post-market surveillance summary, recall record, and the conditions of approval that bound real-world use.
AI Algorithm Version History
ProPro unlocks the chronological record of algorithm version changes — what changed when, drawn from manufacturer changelogs and regulatory filings.
Regulatory Approvals
K202300
Class II
Safety Record
No recalls, field safety notices or FDA safety communications attributable to the Optellum Virtual Nodule Clinic were found in indexed public sources as of July 2026 — read as none found rather than an exhaustive audit, since the FDA recall and MAUDE databases could not be queried directly from this environment. The clinically relevant caution is intrinsic to the indication: the software characterises nodules that have already been identified and is an adjunct to interpretation, not a detector and not an autonomous diagnosis. A low Lung Cancer Prediction score is decision support, not a clearance of malignancy, and the calibrated probability is validated on the nodule populations studied — solid indeterminate nodules within defined size ranges — so it should not be read outside that scope without clinician judgement.
Intended Use & Indications
The Virtual Nodule Clinic is post-processing software that analyses chest CT studies and returns a Lung Cancer Prediction score for indeterminate pulmonary nodules — a calibrated estimate of the probability that a given nodule is malignant. In the US it is cleared as computer-assisted diagnostic software for lesions suspicious of cancer (product code POK, 21 CFR 892.2060, Class II with special controls), an adjunct to the interpreting clinician's own read rather than an autonomous diagnosis. The score is intended to be used alongside a clinician's assessment and established guideline pathways — the British Thoracic Society and Fleischner protocols for incidental nodules, Lung-RADS for screen-detected ones — to help decide between CT surveillance, further imaging, biopsy and surgery. It characterises nodules that have already been identified; it is not a lung-nodule detector and does not screen the whole scan for missed lesions.